Multidimensional infrared spectroscopy is a robust tool for studying the structural dynamics of molecules. In particular, two-dimensional infrared (2DIR) spectroscopy can reveal vibrational coupling among the internal modes of molecules, uncovering the transient structure of complex systems. While spectroscopically very powerful, current experimental techniques are time consuming to perform, requiring ~ 106 laser shots for a single 2DIR spectrum. In this work, we demonstrate a new technique that can acquire a full 2DIR correlation spectrum using a single ultrafast laser pulse. This apparatus will allow 2DIR spectroscopy to be extended to systems that were unattainable with previous technology,including, irreversible chemical reactions, rapid flow experiments, or with low repetition rate laser systems.
© 2007 Optical Society of America
Measuring the structural dynamics of complex systems in solution, regardless of the length-scales or timescales involved, is one of the primary goals of the physical sciences. Most of the Ångstrom spatial resolution (e.g. multi-dimensional NMR spectroscopy). Two-dimensional spectroscopy has shown promise to bridge this gap by providing a local structural probe with picosecond time resolution[1, 2, 3, 4, 5, 6].This spectroscopy has been divided into two general spectral regions: two-dimensional (2D) optical spectroscopy (sensitive to electronic transitions) and two-dimensional Infrared (2DIR)spectroscopy (sensitive to vibrational transitions). Unfortunately, current experimental techniques require ~ 106 laser interactions to compile a single 2D spectrum, making it difficult to acquire spectra on short lived compounds or with the use of low repetition rate (<10Hz) laser systems. We have recently demonstrated a simple spectrometer capable of acquire a complete 2D optical spectrum using a single laser pulse . In this letter, we expand the previous work to demonstrate a spectrometer that allows for a complete 2DIR spectrum to be acquired using a single pump-probe interaction.
An optical analog to multidimensional NMR spectroscopy, 2DIR spectroscopy disperses an IR spectrum across two frequency axes to reveal spectral signatures not apparent in a 1D absorption spectrum. For example, in a simple three-level vibrational system, consisting of v=∣0〉, ∣1〉, and ∣2〉 quantum states, 2DIR spectroscopy can reveal spectral diffusion and vibrational anharmonicity inherent to the system. Situated along the diagonal (ω3=ω1) is a single stimulated emission peak which corresponds to the v=∣1〉→∣0〉 transition. Below the diagonal (ω3<ω1) an induced 2-photon absorption peak is seen representing the anharmonically shifted v =∣1〉→∣2〉 transition. Spectral diffusion is revealed through the structure of the 2D lineshape [3, 8, 9] and vibrational anharmonicities are revealed through the frequency shift of the induced 2-photon absorption transition.
In more complex molecular systems, 2DIR spectra can reveal vibrational coupling. Along the diagonal, two sets of peaks are visible corresponding to the three-level vibrational ladder of a localized excitation. Vibrational coupling between the two localized states is revealed through the existence of peaks in the off-diagonal frequency regions (ω3≠ω1). By analyzing the complex spectral lineshapes, 2DIR can reveal subtle changes in bonding and structural conformations of molecular systems[1, 2, 3]. Given the relatively high frequencies of molecular vibrations (>10THz), the time-resolution of 2DIR can be sub-ps, providing both very high time and localized spatial resolutions.
Two general classes of 2DIR spectroscopy have been developed: Heterodyned Fourier transform 2DIR spectroscopy and double resonance 2DIR spectroscopy[3, 4]. While experimentally distinct, both methods can yield similar information on vibrational coupling and molecular dynamics.Fourier transform 2D spectroscopy uses four interferometrically stabilized laser pulses.Varying the time ordering of the pulses and performing a 2D Fourier transform yields a complex 2D frequency correlation map. The real part of the 2D spectrum corresponds to the absorptive component. This technique provides the highest time-resolution (~50fs), but it requires significant time and experimental resources to acquire the data. Double resonance spectroscopy removes the experimental complexity by using only two laser pulses, a narrow band pump (ω 1)and a broadband probe (ω 3). The narrow band pump pulse selectively excites specific resonances while the probe pulse measures the redistribution of energy to other localized vibrations.The 2D correlation map is generated by slowly stepping the pump frequency while collecting the transient spectra of the probe pulse, resulting in a 2D spectrum that is closely related to the absorptive Fourier transform spectra. The reduction in experimental complexity, however,limits the inherent time resolution to that of the pump pulse (typically a few ps). A significant hindrance to both spectroscopic techniques is that they typically require many laser shots to acquire an single 2D surface. This laser requirement limits the classes of systems that transient 2DIR can study to that of reversible structural transformations and long lived materials.
We have recently developed a 2D spectrometer capable of acquiring 2D optical spectra in a single shot. Based upon double resonance spectroscopy, a frequency dispersed pump pulse excites a system, spatially encoding the pump frequency in the sample. A cylindrically focused probe pulse interrogates the excitation area and is frequency dispersed after the sample (along an axis perpendicular to the pump excitation) to reveal the energy redistribution. Using standard imaging and dispersive optics, the pump and probe frequency axes are recorded onto a 2D array detector. Due to the limited spectral response of charge coupled device (CCD) detectors and extreme cost of IR array detectors (>$100k), the spectrometer was restricted to taking spectra in the near-IR to visible spectral ranges. In this work, we expand our previous work to demonstrate a new 2DIR spectrometer using upconversion IR imaging techniques. To demonstrate the functionality of this spectrometer, we acquire 2DIR spectra of several prototypical compounds in solution as well as demonstrating, for the first time, single-shot 2DIR spectra on an localized vibrational transition.
2. 2DIR spectrometer
The 2DIR spectrometer is a combination of the 2D spectrometer and an upconversion multichannel IR spectrometer (see Fig. 1). The general design of the 2D spectrometer has been discussed in detail previously and will be described briefly here.
The construction of the 2D spectrometer is based upon a dispersed pump-probe setup utilizing two orthogonally oriented grating spectrometers back-to-back. The initial beamsplitter (BS1) is a CaF2 window coated with a dielectric 800nm high-reflection (HR) coating. This window serves two distinct purposes: a beamsplitter for the mid-IR (reflecting ~10% of IR light for use as a probe pulse) and as a mirror for 800nm light. Two plane ruled aluminium gratings, blazed for 4.8µm and with a groove spacing of , are used as the dispersive optic for the 2D spectrometer. The cylindrical focusing mirror has a focal length of 20cm. The first two spherical mirrors (SL1 and SL2) have focal lengths of 20cm while SL3 has a focal length of 10cm. This last lens effectively demagnifies the spectrum in the Fourier plane of the 2D spectrometer. With these optics, the frequency resolution of the 2D spectrometer is estimated to be ~5cm-1.
As in the 2D optical spectrometer, ideally a 2D array detector would be placed directly at the Fourier plane of the 2D spectrometer. While such devices are commercially available for mid-IR radiation detection, they are typically cost prohibitive for most laboratory applications. As an alternative, we choose to upconvert the IR signal to the visible and use silicon CCD technology to detect the upconverted light. In this process, the upconverted intensity is proportional to the product of the IR and visible pulse intensities. Assuming the visible light is well characterized and stable, the upconverted intensity will be directly proportional to incoming IR signal. Prior IR spectroscopic studies have shown that using upconversion to detect IR spectra is very robust and has noise characteristics similar to conventional direct IR detectors[10, 11, 12, 13].
Prior to the upconversion crystal, we place a second 800nm HR on a CaF2 substrate (BS2) to combine the IR signal with an ultrafast 800nm laser pulse for sum frequency generation (SFG). To insure that the upconverted image contains all of the spectral/spatial content of the 2D spectra, we place the large area MgO:LiNbO3(MLN) crystal at the Fourier plane of the 2D spectrometer. The resultant upconverted field was imaged directly onto a silicon CCD array using a single optical lens. Residual 800nm and 400nm (from self doubling of 800nm) light was removed using several optical filters placed prior to the CCD array (BS3). These filters remove almost 100% of the unwanted visible light while only losing ~25% of the upconverted spectra due to surface reflections and absorption. (It should be noted that while the upconverted light is frequency dispersed, the functionality of 2DIR spectrometer will not be effected. For this application, because the upconversion is used to image the Fourier plane of the spectrometer, the spatial intensity of the upconverted light, rather than the color, dominates the upconversion process.)
For the experiments described below, we used a home built Ti:sapphire amplifier system.The laser system produces 40fs pulses at a 1kHz repetition rate. 2.5μJ, 5μm mid-IR light was generated using a home built optical parametric amplifier (OPA). The generated IR pulse has a pulse length of ~80fs with a useable bandwidth of 250cm-1 (FWHM). For upconversion, an uncoated 15×15×0.1mm3 MLN crystal mixed 20µJ of 800nm light with the mid-IR probe pulse to generate the ~700nm visible light. The 100µm thin crystal easily phasematches the entire IR bandwidth for upconversion.
To increase the upconversion efficiency, the power limited 800nm light was down collimated and provided a fluence of ~ at the MLN crystal surface. The upconverted light was collected by a Photometrics Cascade 128+ digital CCD camera. The CCD camera has 128×128 pixels, with a pixel size of 576µm2, and can be triggered at repetition rates greater than 500Hz with 16-bit resolution. Given the observed count rate of visible photons, we estimate the conversion efficiency of the upconversion process to be ~0.1-1%. While this number is relatively small, it was limited only by the very thin upconversion crystal and the available 800nm power.
Due to the limited trigger rate on the CCD camera, the camera was synchronized to the first subharmonic of the laser repetition rate (500Hz). To eliminate exposure to stray visible photons and to ensure that a single laser pulse was collected by the CCD, the exposure time on the camera was set to the minimum 1µs to isolate a single laser pulse. Differential spectra were acquired using an optical chopper, synchronized with the camera, placed in the pump arm at a chopping rate of 250Hz. Every useable laser shot was acquired and subsequent laser shots were subtracted revealing the transient absorption spectra. To increase the signal to noise we performed 2×2 pixel binning, thereby reducing the 128×128 pixel array into 64×64 pixels and increasing the effective pixel size by a factor of 4. After acquiring the signal, 104 static images were collected such that differences in spectral content and upconversion signal intensity could be corrected and that acquired spectra could be normalized.
To calibrate the ω3 axis on the camera, we placed a sample of Rh(CO)2C5H7O2 (RDC) in hexane into the probe arm. RDC has two narrow absorption lines at 2015cm-1and 2084cm-1 corresponding to the symmetric and antisymmetric carbonyl vibrations respectively . The narrow lines allowed the frequency calibration of the pixels in the ω3 direction and verified that the resolution of the spectrometer was ~5cm-1.With the current optics and CCD binning, each pixel corresponds to ~2cm-1 resolution. (For more accurate analysis, the ω3 axis can also be calibrated by spectral interferometry [10, 14]).
To determine temporal and spatial overlap between the pump and probe beams, ∆t=0, we utilized the 800nm HR dielectric coating on BS1. Prior to the spectrometer we overlapped the 800nm and the IR pulses. Placing the MLN crystal at the sample position, the existence of SFG at the MLN crystal verifies the temporal and spatial overlap of the 800nm and IR pulses.Provided that all the optics in the probe arm are reflective, this is an efficient method of determining both spatial and temporal overlap of the pump and probe arms of the 2D spectrometer. The SFG method also serves as an independent verification of the ω1 spectral resolution via the time-inverse of the cross-correlation signal (~4ps).
The resulting SFG signal was also used to crudely calibrate the pump axis. Placing a dense RDC sample in the pump arm, the ∆t=0 measurement becomes an 1D upconversion IR spectrometer. While this does not calibrate theω1 axis on the camera directly, this measurement guarantees that the appropriate pump colors are spatially overlapped with the probe pulse. To calibrate the ω1 axis we used the diffuse scattering of the pump beam from a rough CaF2 window. At t 0, the diffuse scattering will interfere with the probe pulse. Only photons that are identical in both color and location at the sample will interfere, resulting in an interference signal directly along the diagonal at the Fourier plane of the 2D spectrometer. Provided that ω3 axis is calibrated correctly, this interference provides an instant calibration of the ω1 axis. A more rigorous method of calibration can be accomplished by placing a small spatial mask at the sample plane. Measuring the position of the transmitted probe pulse on the CCD camera and correlating that with a independent measurement of the pump frequency.
3. Upconversion 2DIR spectroscopy
To demonstrate the functionality of the 2DIR spectrometer, we measure the 2DIR spectrum of tungsten hexacarbonyl (WHC) and RDC. To eliminate any interference effects between pump and probe beams from diffuse scattering, the spectra were taken at a time delay of ∆t=40 ps. These two prototypical systems demonstrate the two strengths of 2DIR spectroscopy: lineshape analysis and visualizing vibrational coupling.
3.1. 2DIR spectroscopy of WHC
To demonstrate the versatility of the 2DIR spectrometer at observing complex lineshapes we measure the 2D spectra of WHC in hexane and chloroform. The carbonyl stretch of WHC represents a prototypical 3-level system for 2DIR spectroscopy while the two solvents represent a weakly and a strongly interacting solvents. The 1D FTIR spectra of WHC in the two solvents demonstrates the magnitude of the solvent interactions. The carbonyl stretch of WHC in hexane is spectrally narrow (~4cm-1) while the corresponding spectrum in chloroform is 3 times wider and shifted to the red (see Fig. 2).
Spectra of WHC taken with the 2DIR spectrometer shows a 2D spectra of a 3-level system (Fig. 2). Both samples (OD ~0.5) were placed between two 1mm CaF2 windows with a 50µm teflon spacer. 105 of sets of difference spectra were averaged. To eliminate pixel to pixel variation the acquired image was convoluted with a 2D Gaussian whose width (FWHM) was 2 pixels. This width (~4cm-1) is smaller than the measured frequency resolution (~6.5cm-1), thereby preserving all spectral content. In both solvents the stimulated emission signal is clearly seen along the diagonal. Peak differential transmission for the hexane and chloroform solutions is as high as 2.5% and 0.5% respectively. As anticipated from the FTIR, the 2D linewidths and peak positions have shifted appropriately indicating that the 2DIR spectrometer preserves all relevant spectral content. In the off-diagonal regions the induced absorption transition is clearly seen, completing the shape of the prototypical 3-level 2D surface.
Upon closer inspection of the 2DIR spectrum of WHC, we see some differences in the 2D lineshape as a function of solvent. In contrast to the symmetric looking lineshape of the chloroform solution, the hexane sample looks distorted. In particular, the spectral node between the stimulated emission and induced absorption signals is not a flat line, rather it appears that overtone transition is interfering with the fundamental transition. Likewise, above the fundamental transition (ω3 > 1990cm-1 a small induced absorption peak is apparent. This deformation of the 2D lineshape is caused by the high OD and narrow linewidth of the WHC resulting in a relative increase of the induced absorption due to the reabsorption of the stimulated emission signal. The apparent differences between solvents demonstrates the ability of the 2DIR spectrometer to measure complex 2D lineshapes required to see spectral diffusion.
3.2. 2DIR spectroscopy of RDC
To demonstrate the feasibility of measuring cross-peaks using the 2DIR spectrometer, we measure the 2DIR spectrum of RDC in hexane. The symmetric and asymmetric carbonyl vibrations of RDC are anharmonically coupled. The corresponding 2DIR spectra is almost identical to that of an ideal six-level system. Unlike the relatively narrow WHC spectrum, the two transitions of RDC are dispersed across the entire CCD array, requiring a larger area of 800nm light for the upconversion process. To overcome the limited 800nm power, separate 2DIR spectra were taken in different CCD quadrants by moving only the position of the 800nm beam. 105 laser shots were used to acquire each 2D spectrum. The final 2D spectra was compiled by adding the individual difference spectra together (Fig. 3). This method of data acquisition effectively increases the amount of 800nm light (and thus the upconverted intensity) without changing any of the IR optics.
The acquired 2DIR spectrum clearly shows the characteristics of a classic six-level system, two sets of peaks along the diagonal and two sets in the off diagonal spectral regions. The peak change in transmission is 10-4-10-3. Upon further inspection, the relative amplitudes of the peaks are not equal to the anticipated values. In particular the diagonal peaks should be approximately equal and the crosspeaks should be ~1/3 that of the diagonal peaks (assuming that the pump and probe pulses have parallel polarizations). Along each vertical stripe (ω1=constant),however, the ratio of the cross peak to the diagonal peaks are consistent with the expected value, indicating that the individual pump-probe measurements were being performed correctly.
This discrepancy is due to the finite bandwidth and spotsizes of the pump and probe beams at the sample. For example, it is critical that the sample of interest be placed directly at the Fourier plane of the first spectrometer and normal to input direction of the pump beam. Slight deviations in alignment will effect the spectral density in the pump, resulting in differences in the differential probe intensity. The spatial profile of the probe beam can also influence the final 2D spectrum through astigmatic cylindrical focusing at the sample, resulting in poor mode matching of the pump and probe beams. As seen in Fig. 3, these technical misalignments can result in the distortion of the probe intensity as a function of the pump frequency. These effects can be corrected for through careful alignment of the spectrometer and/or data post-processing.
3.3. Single-shot 2DIR spectroscopy
While the 2DIR spectrometer clearly demonstrates the ability to acquire 2D spectra quickly,the real strength is the potential of taking 2DIR spectra in a single shot. In Fig. 4, a series of 2D spectra is shown. Each frame is the average difference spectra between 1, 16, 256, and 4096 pump-probe pairs. It is clear that in a single shot general features of the 2D spectra are apparent, namely a positive emission peak and negative absorption peak in the correct spectral regions as well as spectral node in the correct location. By 256 averages, the general 2D lineshape is visible and by 4096 almost all of the salient features seen in Fig. 2a (distorted 2D line shape, induced absorption peak above positive) are represented in the spectra.
While the spectrometer as demonstrated shows the potential for measuring single-shot 2DIR spectra, several improvements must be made for this spectrometer to work flawlessly throughout the IR spectral region. For example, analysis of the signal-to-noise ratio indicates that the upconversion 2DIR spectrometer is shot-noise limited, resulting in measuring changes as small as 0.1% in under 10 seconds with a 1kHz laser source. In the work, the laser pulse shot-to-shot variability (~ 3%) limited the minimum single-shot resolution.With improvements to the laser stability, the 16-bit camera should be able to measure changes on the order of 10-4 (0.1mOD) with a single pump-probe interaction, sufficient to measure the RDC spectrum. However, to measure the entire RDC spectrum in a single shot, the 800nm power used for upconversion will need to be increased by an order of magnitude to cover the entire desired spectral content. This will also be required for compounds with smaller transition dipole moments, as the non-linear polarization is proportional to µ4. Other improvements include modifying the upconversion techniques for use in the biological ‘fingerprint’ region of the IR spectrum, 6-8µm. This will require either changing the nonlinear crystal used for upconversion (which may require a upconversion laser at a color other than 800nm) or using thin MLN crystals to reduce the mid-IR absorption in the nonlinear crystal. While there are clearly several improvements yet to be made, this spectrometer is a clear first step toward measuring single-shot 2DIR spectra on all potential vibrational systems.
In this paper, we have demonstrated a new technique for the acquisition of 2DIR spectrum with a single laser pulse. The speed and versatility of the technique allows for a large range of experimental systems to now be accessed. Single-shot data acquisition opens the door to measuring 2DIR spectra of short lived molecular species, non-reversible chemical reactions, and relaxes equipment constraints such that low repetition rate laser systems can be used. The speed and lack of moving parts also makes the 2DIR spectrometer ideal for use in rapid screening applications.With modest optical modifications, this spectrometer can be adapted to perform multi-color (using different gratings) and anisotropy measurements (incorporating polarization optics), making it a highly versatile tool for acquiring 2DIR spectra on a wide range molecular species, including biological and organic compounds.
This project received support from the Basic Energy Sciences of the U.S. Department of Energy (Grant# DE-FG02-99ER14988) and the National Science Foundation.
References and links
1. P. Hamm, M. Lim, and R.M. Hochstrasser, “The structure of the amide I band of peptides measured by femtosecond nonlinear IR spectroscopy,” J. Phys. Chem. B ,1026123 (1998). [CrossRef]
2. J.B. Asbury, T. Steinel, and M.D. Fayer, “Using ultrafast Infraed multidimensional correlation spectroscopy to aid in vibrational spectral peak assignments,” Chem. Phys. Lett. 381139 (2003). [CrossRef]
3. M. Khalil, N. Demirdoven, and A. Tokmakoff, “Coherent 2D IR spectoscopy: Molecular Structure and Dynamics in Solution,” J. Phys. Chem. A 1075258 (2003). [CrossRef]
4. V. Cervetto, J. Helbing, J. Bredenbeck, and P. Hamm, “Double-resonance versus pulsed Fourier transform two-dimensional infrared spectroscopy: An experimental and theoretical comparison,” J. Chem. Phys. ,1215935 (2004). [CrossRef] [PubMed]
5. T. Brixner, J. Stenger, H. M. Vaswani, M. Cho, R.E. Blankenship, and G.R. Fleming, “Two-dimensional spectroscopy of electronic couplings in photosynthesis,” Nature 434625 (2005). [CrossRef] [PubMed]
6. C.N. Borca, T. Zhang, X. Li, and S.T. Cundiff, “Optical two-dimensional Fourier transform spectrscopy of semiconductors,” Chem. Phys. Lett. ,416311 (2005). [CrossRef]
8. A. Tokmakoff, “Two-dimensional line shapes derived from coherent third-order nonlinear spectroscopy,” J. Phys. Chem. A ,1044247 (2000). [CrossRef]
9. J.D Hybl, A. Yu, D.A. Farrow, and D. M. Jonas, D. M., “Polar solvation dynamics in the femtosecond evolution of two-dimensional Fourier transform spectra,” J. Phys. Chem. A 1067651 (2002). [CrossRef]
13. K.J. Kubarych, M. Joffre, A. Moore, N. Belabas, and D.M. Jonas “Mid-infrared electric field characterization using a visible charge-coupled-device-based spectrometer,” Opt. Lett. 301228 (2005). [CrossRef] [PubMed]
15. Preliminary experiments have indicated that a 100µm thick piece of MgO:LiNbO3 is partially transparent in the mid-IR spectral regions making upconverion at wavelengths greater than 6µm possible.